
Red flags explained
What a red flag on the page of an exchange or a coin means, which data it is measured from, and what it does not say.
What a red flag is
A red flag is a measurement that does not fit what an exchange or a coin reports about itself, stated with its own figures. It is produced by published rules, the same for everybody, and nobody adds, removes or edits one by hand. It is not a statement that anybody committed fraud, and it changes no price and no score on this site.
Flags are separate from Fraud Signals, which hold documented cases entered by an editor with a link to evidence.
Where the data comes from
Only from what each exchange publishes itself through its public market-data interface, read directly and continuously: prices and 24-hour volumes of every market about once a minute, order books of markets trading at least $100 a day, and the feed of recent public trades about every half hour. No figure comes from another data provider, from a rating, from the press or from user reports.
The measurements themselves are on each exchange's page (order-book backing, estimated real volume, price agreement, data availability) and on each market's row, so every flag can be traced to the numbers under it. The exact formulas and limits are in the methodology.
When a flag appears and disappears
Conditions are evaluated once an hour. A flag is raised only after its condition has held six hours in a row, so a single bad hour, at the exchange or in the reading, raises nothing. It is lowered after six hours in a row without the condition. Exchanges are judged from $1M of measured volume and coins from $1M of daily volume: below that a single trade decides the figures.
Flags on exchanges
Volume without order-book depth
On markets carrying a quarter or more of the exchange's judged volume, the orders resting within 2% of the best bid and of the best ask are worth less than 1% of the volume reported for 24 hours. Only markets trading $10k or more a day are judged.
A market that really trades a billion dollars a day normally shows far more than a few million of orders near the price. On regulated exchanges 95% of markets trade less than 18 times their depth.
Fast market makers can turn over a small book many times; some exchanges publish only part of the book (those are not judged); a one-day burst of trading raises volume before depth follows. Pairs of two dollar-pegged assets are judged only beyond 1,000 times their book.
Trades inside the spread
On markets carrying a quarter or more of judged volume, more than 80% of recent public trades printed strictly between the best bid and the best ask.
A trade against a visible order happens at the bid or at the ask. Trades that keep printing in between were not matched against the book other participants can see.
Hidden or midpoint orders, block trades reported to the public feed, and the delay between reading the book and reading the trades can all put honest trades inside the spread.
Public trades below reported volume
On markets carrying a quarter or more of judged volume, a full day of readings of the public trade feed adds up to less than a fifth of the 24-hour volume the exchange reports.
The trade feed and the volume figure describe the same trading. When they differ fivefold, one of them does not describe what happened.
The reported figure may include trades the public feed leaves out (block or broker trades, other products under the same ticker), or the feed may be cut short by the exchange.
No recent trades
On markets reporting at least $1M a day and carrying a quarter or more of judged volume, the newest public trade is more than 30 minutes old; or, on markets of any size, more than a day old.
A million dollars a day is a trade every few seconds on average. Half an hour of silence does not fit that. Such a market is also left out of the coin's price while other markets of the coin trade: its price is that of a trade long past.
A public trade feed that lags or is paused during maintenance looks the same from outside.
Prices off the market
Over the exchange's markets of coins that other exchanges also trade, with at least $100 traded in 24 hours, the median distance from the Coinsprey price is above 1% and above the exchange's own typical bid/ask spread (above 2% where its tickers publish no spread). Markets that are a different coin under the same ticker are left out.
Where money can move freely, the same coin costs nearly the same everywhere. A persistent gap means prices there are not connected to the rest of the market.
Local-currency exchanges can carry a real premium, deposits or withdrawals may be restricted, and an error on Coinsprey's side produces the same picture: a wrong currency rate and coins sharing a ticker were both found and fixed in October 2026. A gap within the exchange's own spread is shown as "within its spread" and is not flagged.
Data often unavailable
Fewer than 95% of at least 1,000 requests for the exchange's public market data succeeded in 24 hours.
Prices that cannot be read cannot be checked or used.
Maintenance, rate limits applied to the reader, or blocking of its location cause this without anything being wrong at the exchange.
Volume follows another exchange
Over at least 10 markets the exchange shares with one other, larger exchange, its reported volume is the same fraction of that exchange's on every market: the middle half of the ratios lies within 5% of their median. The exchange's page names the other exchange.
Independent trading never lands on one proportion market after market. Volumes that do are computed from the other exchange's figures, and the copied order book and trade feed agree with them, so no other check sees it. Such an exchange is excluded from price calculation.
An exchange that passes its customers' orders on to a larger one, or a broker on its liquidity, reports that exchange's trading, in full or in part; that is routing, not invention, but the same trades are then already counted under the larger exchange.
Two different volumes published
The exchange publishes its markets in two public lists: its trading API and a second list, usually one made for price trackers. Over at least 10 markets found in both (each reporting $10,000 a day or more), the typical market's 24-hour volume in one list is at least 1.5 times its volume in the other. Both lists are read from the exchange itself, at the same time.
One market has one traded volume in 24 hours. An exchange that gives two answers shows that at least one of them is not a count of trades, and it explains why the same exchange stands at different volumes on different trackers. The figure shown on Coinsprey is the one from the trading API.
One list may count both sides of each trade and the other one side, the two may cover different periods, or one may leave out some kinds of orders. Each of these is a matter of definition, not invention, but the exchange does not say which applies.
Volume without visitors
The exchange reports $100M or more a day, and its website is not among the million most visited websites in the Chrome UX Report, a ranking Google publishes monthly from the browsing of Chrome users.
Volume, order book and trade feed are all published by the exchange itself and can be made to agree with each other. Visitors cannot: an exchange that turns over more than well-known ones is expected to have a website people open.
The ranking counts website visits in Chrome only. Customers who trade through a mobile app or through the API, institutional venues with few accounts, and exchanges serving countries where Chrome is rare all leave little trace in it. The ranking gives a band, not an exact place, and is a month old at most.
Flags on coins
Priced on one exchange
A single exchange forms the coin's price, on at least $1M of daily volume.
With one venue there is nothing to compare the price with, and that venue's outage or error is the coin's price.
New coins usually start on one exchange. The flag describes concentration, not the coin's quality.
Priced only by exchanges not included in prices
Every exchange trading the coin fails a check that would let its markets count (prices agree, data arrive, order books readable).
The price shown rests only on figures that could not be verified against anything.
An exchange may fail a check for a technical reason, for example because it publishes no order book.
Low price confidence
The price confidence score is below 40 of 100.
Few venues, thin books, wide spreads or disagreeing prices: the number shown is a weak estimate.
The components of the score are on the coin's page and say which of these it is.
Reported volume mostly contradicted
Over judged markets carrying at least half of the coin's volume, the estimate of real volume is below a quarter of what is reported.
Most of the trading reported for the coin sits on markets where a measurement contradicts it.
The same alternative explanations apply as for the exchange flags above; the estimate is deliberately cut only on hard contradictions.
What a flag does not say
It does not say why the measurement looks the way it does, and several ordinary explanations are listed above for each. It says nothing about custody, solvency, licences or the people behind an exchange: none of that is measured here. An exchange or a coin without a flag is not thereby safe; a large market with a deep book and trades at the best prices contradicts nothing, inflated or not.
If a flag is wrong
Measurements can be wrong: an interface changes, a book is published in part, a rate is off. An exchange or a project that believes a flag rests on such an error can describe it through the report form, with the market concerned. Where the measurement is at fault the rule or the reader is corrected for everybody and the change is noted in the methodology; flags are not removed on request or for payment.


